danium/lateral-thinking

lateral‑thinking

What it is

• A collection of skills (plug‑in modules) for AI coding agents that give the agent structured creative‑thinking techniques (random‑stimulus, provocation, inversion, SCAMPER, Six‑Thinking‑Hats, etc.).
• Each skill implements a classic “lateral‑thinking” method, adds an abandonment rule (when to give up) and makes any failures visible, so the user can see why some prompts produced no ideas.
• A small router (lateral) diagnoses the symptom you describe (e.g., “ideas feel predictable”) and selects the appropriate technique.

Why it matters

• Helps AI‑assisted product teams break out of obvious feature ideas and explore alternative product directions.
• By surfacing empty‑operation results, it turns “nothing came up” into a useful insight about the problem space.
• Works with any agent that supports the skill interface (Claude Code, Codex, generic “npx skills” agents, etc.).

Installation

# generic agent (recommended)
npx skills add danium/lateral-thinking

# Claude Code plugin
/plugin marketplace add danium/lateral-thinking
/plugin install lateral-thinking@lateral-thinking

# Codex
$skill‑installer https://github.com/danium/lateral-thinking

# Manual
git clone … && copy skills/* into your agent’s skills folder

The install drops a router (lateral) and a set of sibling technique skills.

How to use

1. Ask the agent to use the lateral skill or explicitly invoke a technique, e.g.

   agent: use lateral skill
   

2. The router asks you to describe the symptom (e.g., “ideas all feel the same”).
3. It selects the matching technique (random‑stimulus, provocation, inversion, …) and runs it.
4. The technique generates up to seven ideas, marking any empty operations and applying its abandonment rule.
5. After the batch, the router prompts you to look for a meta‑pattern—the structural insight that explains why certain operations failed.

Demo snippet (from the README)

• Prompt: “Our users create one project, then never return. What product direction would make that first project worth coming back to?”
• Vanilla agent returns three generic feature ideas.
• Agent with random‑stimulus produces richer concepts like a “lighthouse beam” return trigger and a “sourdough starter” weekly brief, illustrating how the technique nudges the model toward novel signals rather than reminders.

Supported techniques

Symptom	Technique
Predictable brainstorm	random-stimulus
Unbreakable constraint	provocation
Hidden assumptions	inversion
Wrong problem focus	concept-fan
Derivative solution	analogy
Need variations on one idea	scamper
Too‑fast consensus	six-hats
Overly safe thinking	worst-idea

What it is not

• Not a magic “prompt pack” that simply adds more text to a model.
• Not a standalone product; it requires an AI coding agent that can load and run skills.

License: MIT.

---

Bottom line: lateral‑thinking is a lightweight, open‑source toolkit that equips AI agents with proven creative‑thinking frameworks and explicit failure reporting, enabling teams to surface non‑obvious product directions directly from their coding assistants.